How does protein interaction detection method depend on the protein length?
Vitalii Kleshchevnikov
13 December 2016
setting up parameters of the analysis
=========================================================
Species to select from: “Homo sapiens”, “Mus musculus”, “strain ATCC 204508”, “strain K12” Paste value in the code chunk below
SPECIES_NAME = "Homo sapiens"all Uniprot if reviewed == 1, only Swissprot data if reviewed == 2, TrEMBL data if reviewed == 3 only reviewed = 2 is relevant for this analysis Paste value in the code chunk below
reviewed = 2To distinguish between isoforms or not (to use only generic UniprotACs): TRUE / FALSE? only isoforms = FALSE is relevant for this analysis Paste value in the code chunk below
isoforms = FALSEPlease specify the date for which you want to perform the analysis (if not today) “Logic table” from protein_properties script is needed for this analysis. Filename example:
“proteome_vs_interactome_protein_properties_f_Homo sapiens_reviewed_2_isoforms_FALSE_2016-12-01.txt”
date = Sys.Date()
date = as.Date("2016-12-01")the date of analysis, how old is Uniprot protein list: 2016-12-01
Species ID lookup
=========================================================
looking up species ID in the Uniprot readme file (always reads from url)
source("SPECIES_NAME_TO_ID.R")
r = reviewed
i = isoforms
n = SPECIES_NAME
SPECIES_IDs = SPECIES_NAME_TO_ID(n)## [1] "UP000005640 9606 HUMAN 21032 71899 93381 Homo sapiens (Human)"
## Proteome_ID SPECIES_ID
## 1 UP000005640 9606SPECIES_ID = SPECIES_IDs$SPECIES_IDquerying for interations detected by AP-MS or two hybrid using PSICQUIC
=========================================================
Detection method explanation:
- detmethod:MI:0232 - transcriptional complementation assay, includes different two-hybrids http://www.ebi.ac.uk/ols/ontologies/mi/terms?iri=http%3A%2F%2Fpurl.obolibrary.org%2Fobo%2FMI_0232
detmethod:MI:0090 - protein complementation assay, includes transcriptional complementation assay and bimolecular fluorescence complementation, enzyme complementations http://www.ebi.ac.uk/ols/ontologies/mi/terms?iri=http%3A%2F%2Fpurl.obolibrary.org%2Fobo%2FMI_0090
- both detmethod and pmethod are required for specifying AP-MS
- detmethod:MI:0004 - affinity chromatography technology, co-IP and pulldowns http://www.ebi.ac.uk/ols/ontologies/mi/terms?iri=http%3A%2F%2Fpurl.obolibrary.org%2Fobo%2FMI_0004
pmethod:MI:0433 - partial identification of protein sequence http://www.ebi.ac.uk/ols/ontologies/mi/terms?iri=http%3A%2F%2Fpurl.obolibrary.org%2Fobo%2FMI_0433
to change: type in exact database names (the list below - all IMEx -is default for the function)
databases <- c("IntAct", "MINT", "bhf-ucl", "MPIDB", "MatrixDB",
"HPIDb","I2D-IMEx","InnateDB-IMEx", "MolCon", "UniProt", "MBInfo")
source("query_PSICQUIC_for_interactions.R")
twohybrids_all_interactions = query_PSICQUIC_for_interactions(SPECIES_ID = SPECIES_ID,
SPECIES_NAME = SPECIES_NAME,
databases = databases, date,
detmethod = "transcriptional complementation assay")## [1] "loaded from file"
## [1] interactions for Homo sapiens, detmethod(transcriptional complementation assay), pmethod(NA):
## [1] total number: 135332
## [1] there is no interactions in the databases:
## [1] MPIDB MatrixDB MBInfo
## [1] the number of interactions per database
## database N of interactions
## 1 53
## 2 psi-mi:MI:0469(IntAct) 123945
## 3 psi-mi:MI:0471(MINT) 9834
## 4 psi-mi:MI:0486(UniProt) 771
## 5 psi-mi:MI:0903(mpidb) 3
## 6 psi-mi:MI:1222(mbinfo) 59
## 7 psi-mi:MI:1262(I2D) 39
## 8 psi-mi:MI:1263(Molecular Connections) 41
## 9 psi-mi:MI:1332(bhf-ucl) 83
## 10 psi-mi:MI:1335(HPIDb) 504complementation_all_interactions = query_PSICQUIC_for_interactions(SPECIES_ID = SPECIES_ID,
SPECIES_NAME = SPECIES_NAME,
databases = databases, date,
detmethod = "protein complementation assay")## [1] "loaded from file"
## [1] interactions for Homo sapiens, detmethod(protein complementation assay), pmethod(NA):
## [1] total number: 139285
## [1] there is no interactions in the databases:
## [1] MPIDB MatrixDB MBInfo
## [1] the number of interactions per database
## database N of interactions
## 1 54
## 2 psi-mi:MI:0469(IntAct) 126515
## 3 psi-mi:MI:0471(MINT) 10329
## 4 psi-mi:MI:0486(UniProt) 877
## 5 psi-mi:MI:0903(mpidb) 3
## 6 psi-mi:MI:0974(InnateDB) 21
## 7 psi-mi:MI:1222(mbinfo) 66
## 8 psi-mi:MI:1262(I2D) 87
## 9 psi-mi:MI:1263(Molecular Connections) 50
## 10 psi-mi:MI:1332(bhf-ucl) 722
## 11 psi-mi:MI:1335(HPIDb) 561ap_ms_all_interactions = query_PSICQUIC_for_interactions(SPECIES_ID = SPECIES_ID,
SPECIES_NAME = SPECIES_NAME,
databases = databases, date,
detmethod = "affinity chromatography technology",
pmethod = "partial identification of protein sequence")## [1] "loaded from file"
## [1] interactions for Homo sapiens, detmethod(affinity chromatography technology), pmethod(partial identification of protein sequence):
## [1] total number: 88410
## [1] there is no interactions in the databases:
## [1] MPIDB MatrixDB MBInfo
## [1] the number of interactions per database
## database N of interactions
## 1 92
## 2 psi-mi:MI:0469(IntAct) 85442
## 3 psi-mi:MI:0471(MINT) 865
## 4 psi-mi:MI:0486(UniProt) 1241
## 5 psi-mi:MI:0974(InnateDB) 2
## 6 psi-mi:MI:1262(I2D) 72
## 7 psi-mi:MI:1263(Molecular Connections) 79
## 8 psi-mi:MI:1332(bhf-ucl) 47
## 9 psi-mi:MI:1335(HPIDb) 570Cleaning and transforming data
=========================================================
extracting interactor ID-s from interactions (MI-TAB 2.5)
source("interactions_to_interactors.R")
twohybrids_all_interactors = interactions_to_interactors(twohybrids_all_interactions)
complementation_all_interactors = interactions_to_interactors(complementation_all_interactions)
ap_ms_all_interactors = interactions_to_interactors(ap_ms_all_interactions)filtering interactors for uniprotkb only indentifiers filtering for SPECIES_ID only proteins
source("uniprotkb_and_SPECIES_ID_interactor_selector.R")
twohybrids_all_interactors_SPECIES_ID_only = uniprotkb_and_SPECIES_ID_interactor_selector(twohybrids_all_interactors, SPECIES_ID)## all interactors
## "18408"
## interactors with the UniprotKB identifier
## "17969"
## interactors with the other identifier
## "439"
## SPECIES_ID
## "9606"
## SPECIES_ID interactors
## "13933"
## interactors from other species
## "4060"complementation_all_interactors_SPECIES_ID_only = uniprotkb_and_SPECIES_ID_interactor_selector(complementation_all_interactors, SPECIES_ID)## all interactors
## "18985"
## interactors with the UniprotKB identifier
## "18479"
## interactors with the other identifier
## "506"
## SPECIES_ID
## "9606"
## SPECIES_ID interactors
## "14319"
## interactors from other species
## "4185"ap_ms_all_interactors_SPECIES_ID_only = uniprotkb_and_SPECIES_ID_interactor_selector(ap_ms_all_interactors, SPECIES_ID)## all interactors
## "13375"
## interactors with the UniprotKB identifier
## "12456"
## interactors with the other identifier
## "919"
## SPECIES_ID
## "9606"
## SPECIES_ID interactors
## "10766"
## interactors from other species
## "1732"Removing all isoform IDs (XXXXXX-X+ => XXXXXX) from IDs
source("isoform_id_all_remover.R")
twohybrids_all_interactors_SPECIES_ID_only$interactor_IDs = isoform_id_all_remover(twohybrids_all_interactors_SPECIES_ID_only$interactor_IDs)
complementation_all_interactors_SPECIES_ID_only$interactor_IDs = isoform_id_all_remover(complementation_all_interactors_SPECIES_ID_only$interactor_IDs)
ap_ms_all_interactors_SPECIES_ID_only$interactor_IDs = isoform_id_all_remover(ap_ms_all_interactors_SPECIES_ID_only$interactor_IDs)preparing interaction detection method data for logic table: selecting unique proteins and adding the column of ones
unique_twohybrids_interactors_SPECIES_ID_only = unique(cbind(twohybrids_all_interactors_SPECIES_ID_only[c("interactor_IDs")], 1))
colnames(unique_twohybrids_interactors_SPECIES_ID_only)[2] = "two_hybrid"
unique_complementation_interactors_SPECIES_ID_only = unique(cbind(complementation_all_interactors_SPECIES_ID_only[c("interactor_IDs")], 1))
colnames(unique_complementation_interactors_SPECIES_ID_only)[2] = "all_protein_complementation"
unique_ap_ms_interactors_SPECIES_ID_only = unique(cbind(ap_ms_all_interactors_SPECIES_ID_only[c("interactor_IDs")], 1))
colnames(unique_ap_ms_interactors_SPECIES_ID_only)[2] = "AP_MS"loading logic table made by “swissprot_vs_imex_protein_properties” script
filename_vs_2 = paste("./analysis/","proteome_vs_interactome_protein_properties_f_", n,"_reviewed_",r,"_isoforms_",i,"_", date,".txt", sep = "")
proteome_vs_imex_details_f = as.data.frame(read.delim(filename_vs_2, header = T, stringsAsFactors = F,quote=""))
proteome_vs_imex_details_f$whole_proteome_Uniprot_IMEx = factor(proteome_vs_imex_details_f$whole_proteome_Uniprot_IMEx, ordered =F)merging new results with the logic table
proteome_vs_imex_interaction_details_t1 = merge(proteome_vs_imex_details_f,
unique_twohybrids_interactors_SPECIES_ID_only,
by.x = "whole_proteome_IDs",
by.y = "interactor_IDs",
all.x = T, all.y = F)
proteome_vs_imex_interaction_details_t2 = merge(proteome_vs_imex_interaction_details_t1,
unique_complementation_interactors_SPECIES_ID_only,
by.x = "whole_proteome_IDs",
by.y = "interactor_IDs",
all.x = T, all.y = F)
proteome_vs_imex_interaction_details_f = merge(proteome_vs_imex_interaction_details_t2,
unique_ap_ms_interactors_SPECIES_ID_only,
by.x = "whole_proteome_IDs",
by.y = "interactor_IDs",
all.x = T, all.y = F)
proteome_vs_imex_interaction_details_f[is.na(proteome_vs_imex_interaction_details_f)] = 0adding factor combination variable - two_hybrid and AP_MS
proteome_vs_imex_interaction_details_f[,length(proteome_vs_imex_interaction_details_f)+1] = interaction(proteome_vs_imex_interaction_details_f$two_hybrid, proteome_vs_imex_interaction_details_f$AP_MS, sep = "_")
colnames(proteome_vs_imex_interaction_details_f)[length(proteome_vs_imex_interaction_details_f)] = paste0("two_hybrid", "_vs_","AP_MS")
levels(proteome_vs_imex_interaction_details_f$two_hybrid_vs_AP_MS) = c("not_two_hybrid_and_not_AP_MS", "two_hybrid_not_AP_MS","not_two_hybrid_but_AP_MS", "two_hybrid_and_AP_MS")
proteome_vs_imex_interaction_details_f$two_hybrid_vs_AP_MS = as.character(proteome_vs_imex_interaction_details_f$two_hybrid_vs_AP_MS)
proteome_vs_imex_interaction_details_f$two_hybrid_vs_AP_MS[proteome_vs_imex_interaction_details_f$IMEx!=1] = "not_in_IMEx"saving combined logic table with protein properties from Uniprot and interaction properties combined
filename = paste("./analysis/","proteome_vs_interactome_interaction_properties_f_", n,"_reviewed_",r,"_isoforms_",i,"_", date,".txt", sep = "")
write.table(proteome_vs_imex_interaction_details_f,filename,col.names=T,row.names=F,sep="\t",quote=F)Results
=========================================================
read the table saved before - script can be started from here
filename = paste("./analysis/","proteome_vs_interactome_interaction_properties_f_", n,"_reviewed_",r,"_isoforms_",i,"_", date,".txt", sep = "")
proteome_vs_imex_interaction_details_f = as.data.frame(read.delim(filename, header = T, stringsAsFactors = F,quote=""))
proteome_vs_imex_interaction_details_f$whole_proteome_Uniprot_IMEx = factor(proteome_vs_imex_interaction_details_f$whole_proteome_Uniprot_IMEx, ordered =F)
proteome_vs_imex_interaction_details_f$two_hybrid_vs_AP_MS = factor(proteome_vs_imex_interaction_details_f$two_hybrid_vs_AP_MS, ordered =F)The density and the histogram of the protein mass (overlay for different detection methods)
=========================================================
library(ggplot2)
ggplot(proteome_vs_imex_interaction_details_f, aes(x = Mass, color =two_hybrid_vs_AP_MS,
fill = two_hybrid_vs_AP_MS)) +
geom_density(alpha =0.2) + scale_x_log10() +
xlab("protein mass, Da, log10 scale")+
ggtitle("density plot of the protein mass (overlay for different detection methods)") #+facet_grid(two_hybrid_vs_AP_MS~Organism)
ggplot(proteome_vs_imex_interaction_details_f, aes(x = Mass,color = two_hybrid_vs_AP_MS)) +
scale_x_log10() + geom_histogram(position = "identity", bins = 50,alpha =0.1) +
xlab("protein mass, Da, log10 scale") +
ggtitle("histogram of the protein mass (overlay for different detection methods)")
#+facet_grid(two_hybrid_vs_AP_MS~Organism)How does protein interaction detection method depend on the protein length? - Violin plot
library(dplyr)##
## Attaching package: 'dplyr'## The following objects are masked from 'package:stats':
##
## filter, lag## The following objects are masked from 'package:base':
##
## intersect, setdiff, setequal, unionlibrary(scales)
## adding two_hybrid and AP_MS levels to show non-combinatory relationship (not
## excluding proteins present in both)
two_hybrid = filter(proteome_vs_imex_interaction_details_f, two_hybrid == 1)
two_hybrid$two_hybrid_vs_AP_MS = "two_hybrid"
AP_MS = filter(proteome_vs_imex_interaction_details_f, AP_MS == 1)
AP_MS$two_hybrid_vs_AP_MS = "AP_MS"
for_plot = rbind(proteome_vs_imex_interaction_details_f, two_hybrid, AP_MS)
for_plot$two_hybrid_vs_AP_MS = factor(for_plot$two_hybrid_vs_AP_MS, ordered = F)
## Calculating protein median protein mass for each group
yy = split(for_plot, for_plot$two_hybrid_vs_AP_MS)
Mass_median = sapply(yy, function(x){median(log10(x$Mass))})
kDa_median_mass = (10^Mass_median)/1000
plot_labels = paste0(gsub("_"," ",levels(for_plot$two_hybrid_vs_AP_MS)),", \n (median mass: ", signif(kDa_median_mass,3)," kDa)")
## Violin plot - proportional
ggplot(for_plot, aes(y = Mass, x =two_hybrid_vs_AP_MS, fill = two_hybrid_vs_AP_MS)) +
geom_violin(draw_quantiles = c(0.05,0.25,0.495,0.5,0.505,0.75,0.95),scale = "count", alpha =0.7) +
scale_y_log10(breaks = trans_breaks("log10", function(x) 10^x),
labels = trans_format("log10", math_format(10^.x))) +
scale_x_discrete(labels = plot_labels)+
geom_abline(slope = 0, intercept = Mass_median, alpha =0.1) +
ylab("Mass, Da, log10 scale")+
xlab("presence in IMEx, detection method")+
ggtitle("How does protein interaction detection method depend on the protein length?",
subtitle = "violin areas are scaled proportionally to the number of observations")+
coord_flip()#+facet_grid(two_hybrid_vs_AP_MS~Organism)
## Violin plot - not proportional
ggplot(for_plot, aes(y = Mass, x =two_hybrid_vs_AP_MS, fill = two_hybrid_vs_AP_MS)) +
geom_violin(draw_quantiles = c(0.05,0.25,0.495,0.5,0.505,0.75,0.95),scale = "area", alpha =0.7) +
scale_y_log10(breaks = trans_breaks("log10", function(x) 10^x),
labels = trans_format("log10", math_format(10^.x))) +
scale_x_discrete(labels = plot_labels)+
geom_abline(slope = 0, intercept = Mass_median, alpha =0.1) +
ylab("Mass, Da, log10 scale")+
xlab("presence in IMEx, detection method")+
ggtitle("How does protein interaction detection method depend on the protein length?",
subtitle = "all violins have the same area")+
coord_flip()#+facet_grid(two_hybrid_vs_AP_MS~Organism)
print(signif(data.frame("mass_median(log10.Da)" = Mass_median, "mass_median(kDa)" = (10^Mass_median)/1000),3))## mass_median.log10.Da. mass_median.kDa.
## not_in_IMEx 4.56 36.7
## not_two_hybrid_and_not_AP_MS 4.70 50.0
## not_two_hybrid_but_AP_MS 4.76 57.3
## two_hybrid_and_AP_MS 4.75 56.2
## two_hybrid_not_AP_MS 4.63 42.6
## two_hybrid 4.70 50.0
## AP_MS 4.75 56.6Let’s filter all the proteins composing 36.7kDa peak in non-represented-in-IMEx group
proteome_vs_imex_interaction_details_f.not_in_IMEx = dplyr::filter(
proteome_vs_imex_interaction_details_f, two_hybrid_vs_AP_MS == "not_in_IMEx")
proteome_vs_imex_interaction_details_f.32.42 = dplyr::filter(
proteome_vs_imex_interaction_details_f.not_in_IMEx,
32000 < proteome_vs_imex_interaction_details_f.not_in_IMEx$Mass)
proteome_vs_imex_interaction_details_f.32.42 = dplyr::filter(
proteome_vs_imex_interaction_details_f.32.42,
proteome_vs_imex_interaction_details_f.32.42$Mass < 42000)
#proteome_vs_imex_interaction_details_f.32.42.biogrid = dplyr::filter(
# proteome_vs_imex_interaction_details_f.32.42,
# proteome_vs_imex_interaction_details_f.32.42$BioGRID_from_Mentha == 0)
#proteome_vs_imex_interaction_details_f.32.42.biogrid.annotation5.5 = dplyr::filter(
# proteome_vs_imex_interaction_details_f.32.42.biogrid,
# proteome_vs_imex_interaction_details_f.32.42.biogrid$Annotation == "5 out of 5")
write(proteome_vs_imex_interaction_details_f.32.42$whole_proteome_IDs, "proteome_vs_imex_interaction_details_f.32.42.txt")
#table(proteome_vs_imex_interaction_details_f.not_in_IMEx)
#swissprot = filter(proteome_vs_imex_interaction_details_f,
# proteome_vs_imex_interaction_details_f$whole_proteome_Uniprot == 1)
#write(swissprot$whole_proteome_IDs, "swissprot.txt")Quick GO enrichment in Cytoscape ClueGO showed that large chunk of the missing human interactome consists of odor receptors (36kDa spike), also bitter taste receptors ~20 genes, membrane transporters.
GO terms decribing missing proteins
cell.process = EBImage::readImage('./results/proteome_vs_imex_interaction_details_f.not_in_IMEx.cell.process.png')
EBImage::display(cell.process)
detailed GO enrichment results available here - “./results/proteome_vs_imex_details_f.not_in_IMEx.ClueGO/”
Removed olfactory receptors
## removing olfactory receptors:
proteome_vs_imex_details_f_minus_odor = proteome_vs_imex_interaction_details_f[-grep("Odor", proteome_vs_imex_interaction_details_f$Protein.names),]
proteome_vs_imex_details_f_minus_odor_olf = proteome_vs_imex_details_f_minus_odor[-grep("Olfactory", proteome_vs_imex_details_f_minus_odor$Protein.names),]
print(paste("number of proteins containing \"Odor\" or \"Olfactory\" in protein name: ", length(grep("Odor", proteome_vs_imex_interaction_details_f$Protein.names)) +
length(grep("Olfa", proteome_vs_imex_interaction_details_f$Protein.names))))## [1] "number of proteins containing \"Odor\" or \"Olfactory\" in protein name: 438" ## density without olfactory receptors:
#ggplot(proteome_vs_imex_details_f_minus_odor_olf, aes(x = Mass, color = whole_proteome_Uniprot_IMEx, alpha =0.5)) +geom_density()+ scale_x_log10()
two_hybrid = dplyr::filter(proteome_vs_imex_details_f_minus_odor_olf, two_hybrid == 1)
two_hybrid$two_hybrid_vs_AP_MS = "two_hybrid"
AP_MS = dplyr::filter(proteome_vs_imex_details_f_minus_odor_olf, AP_MS == 1)
AP_MS$two_hybrid_vs_AP_MS = "AP_MS"
for_plot = rbind(proteome_vs_imex_details_f_minus_odor_olf, two_hybrid, AP_MS)
for_plot$two_hybrid_vs_AP_MS = factor(for_plot$two_hybrid_vs_AP_MS, ordered = F)
## Calculating protein median protein mass for each group
yy = split(for_plot, for_plot$two_hybrid_vs_AP_MS)
Mass_median = sapply(yy, function(x){median(log10(x$Mass))})
kDa_median_mass = (10^Mass_median)/1000
plot_labels = paste0(gsub("_"," ",levels(for_plot$two_hybrid_vs_AP_MS)),", \n (median mass: ", signif(kDa_median_mass,3)," kDa)")
## Violin plot - proportional
ggplot(for_plot, aes(y = Mass, x =two_hybrid_vs_AP_MS, fill = two_hybrid_vs_AP_MS)) +
geom_violin(draw_quantiles = c(0.05,0.25,0.495,0.5,0.505,0.75,0.95),scale = "count", alpha =0.7) +
scale_y_log10(breaks = trans_breaks("log10", function(x) 10^x),
labels = trans_format("log10", math_format(10^.x))) +
scale_x_discrete(labels = plot_labels)+
geom_abline(slope = 0, intercept = Mass_median, alpha =0.1) +
ylab("Mass, Da, log10 scale")+
xlab("presence in IMEx, detection method")+
ggtitle("How does protein interaction detection method depend on the protein length?",
subtitle = "all violins have the same area")+
coord_flip()#+facet_grid(two_hybrid_vs_AP_MS~Organism)
## Violin plot - not proportional
ggplot(for_plot, aes(y = Mass, x =two_hybrid_vs_AP_MS, fill = two_hybrid_vs_AP_MS)) +
geom_violin(draw_quantiles = c(0.05,0.25,0.495,0.5,0.505,0.75,0.95),scale = "area", alpha =0.7) +
scale_y_log10(breaks = trans_breaks("log10", function(x) 10^x),
labels = trans_format("log10", math_format(10^.x))) +
scale_x_discrete(labels = plot_labels)+
geom_abline(slope = 0, intercept = Mass_median, alpha =0.1) +
ylab("Mass, Da, log10 scale")+
xlab("presence in IMEx, detection method")+
ggtitle("How does protein interaction detection method depend on the protein length?",
subtitle = "all violins have the same area")+
coord_flip()#+facet_grid(two_hybrid_vs_AP_MS~Organism)
# creating a list of missing proteins for ClueGO
proteome_vs_imex_details_f_minus_odor_olf.not_in_IMEx = dplyr::filter(
proteome_vs_imex_details_f_minus_odor_olf, two_hybrid_vs_AP_MS == "not_in_IMEx")
write(proteome_vs_imex_details_f_minus_odor_olf.not_in_IMEx$whole_proteome_IDs, "proteome_vs_imex_interaction_details_f.not_in_IMEx_olfac_removed.txt")GO terms decribing missing proteins (after removing olfactory receptors)
cell.process.olf.removed = EBImage::readImage('./results/proteome_vs_imex_interaction_details_f.not_in_IMEx.ofl_removed.cell.process.png')
EBImage::display(cell.process.olf.removed)
detailed GO enrichment results available here - “./results/proteome_vs_imex_details_f_minus_odor_olf.not_in_IMEx_ClueGO/”
#image0 = proteome_vs_imex_interaction_details_f[,c(2:3,6:18,29:31)]
#image = as.data.frame(lapply(image0, function(x){as.numeric(x)}))
#rafalib::mypar()
#rafalib::imagemat(image)
#colnames(image)
image1 = proteome_vs_imex_interaction_details_f[,c(1:3,6:18,30:32)]
image1.melt = reshape2::melt(image1)## Using whole_proteome_IDs as id variablescolnames(image1.melt) = c("whole_proteome_IDs", "group", "presence")
ggplot(image1.melt, aes(group, whole_proteome_IDs)) + geom_tile(aes(fill = presence)) + scale_fill_gradient(low = "white", high = "black") + theme(axis.text.x = element_text(size = 8, angle = 10, margin = margin(t=8)))